DOC: document read_csv fastpath for iso8601 datetime strings

[ghost]

While investigating an optimization suggested in #5490 , I came
across a fast path in pd.read_csv that I was unaware of for parsing iso8601 datetimes.
It's undocumented as far as I can tell.

RIght the wrong.

N=10000
r=pd.date_range("2000-01-01","2013-01-01",freq="H")[:N]
df=pd.DataFrame(range(N),index=r)
df.to_csv("/tmp/1.csv")
def f1(s):
    return dt.datetime.strptime(s,"%Y-%m-%d %H:%M:%S")
def f2(s):
    return dateutil.parser.parse(s)

%timeit x1=pd.read_csv("/tmp/1.csv",parse_dates=True,index_col=0) # fast-path for iso8601, fallback to dateutil.parser
%timeit x2=pd.read_csv("/tmp/1.csv",parse_dates=True,index_col=0,date_parser=f1) 
%timeit x2=pd.read_csv("/tmp/1.csv",parse_dates=True,index_col=0,date_parser=f2)

100 loops, best of 3: 14.1 ms per loop
1 loops, best of 3: 259 ms per loop
1 loops, best of 3: 956 ms per loop